Prostaglandin E2 Activates HPK1 Kinase Activity via a PKA-dependent Pathway*

Hematopoietic progenitor kinase 1 (HPK1) is a hematopoietic cell-restricted member of the Ste20 serine/threonine kinase super family. We recently reported that the immunosuppressive eicosanoid, prostaglandin E2 (PGE2), is capable of activating HPK1 in T cells. In this report, we demonstrate that unlike the TCR-induced activation of HPK1 kinase activity, the induction of HPK1 catalytic activity by PGE2 does not require the presence of phosphotyrosine-based signaling molecules such as Lck, ZAP-70, SLP-76, and Lat. Nor does the PGE2-induced HPK1 activation require the intermolecular interaction between its proline-rich regions and the SH3 domain-containing adaptor proteins, as required by the signaling from the TCR to HPK1. Instead, our study reveals that PGE2 signal to HPK1 via a 3′ -5 ′-cyclic adenosine monophosphate-regulated, PKA-dependent pathway. Consistent with this observation, changing the serine 171 residue that forms the optimal PKA phosphorylation site within the “activation loop” of HPK1 to alanine completely prevents this mutant from responding to PGE2-generated stimulation signals. Moreover, the inability of HPK1 to respond to PGE2 stimulation in PKA-deficient S49 cells further supports the importance of PKA in this signaling pathway. We speculate that this unique signaling pathway enables PGE2 signals to engage a proven negative regulator of TCR signal transduction pathway and uses it to inhibit T cell activation.

Hematopoietic progenitor kinase 1 (HPK1) 2 is a member of the GCK sub-family of the Ste20 kinases (1). HPK1 is expressed ubiquitously in all embryonic tissues examined, but this expression profile shifts to a hematopoietic cell-restricted pattern postpartum at neonatal day 1, leading to speculation that it may perform a specialized function in hematopoietic cells (2). The role of HPK1 in biological processes is characterized best in T cells, where HPK1 has emerged as an important negative regulator of T cell antigen receptor (TCR)-induced interleukin-2 gene transcription (3,4). In addition to controlling interleukin-2 gene transcription, overexpression studies indicate that HPK1 also plays a role in activation-induced cell death upon TCR engagement (5,6). Although some of these overexpression studies produced conflicting data as to the role that HPK1 plays in these T cell functions (4,(7)(8)(9), a recent study reveals that T cells from HPK1 Ϫ/Ϫ mice proliferate more robustly in response to TCR engagement (10). HPK1 Ϫ/Ϫ mice also exhibited a more severe autoimmune phenotype in the experimental model of autoimmune encephalomyelitis. These findings firmly validate the earlier biochemical findings that HPK1 functions as a negative regulator of T cell activation.
The catalytic activity of HPK1 is elevated upon ligand engagement of a variety of cell surface receptors present on hematopoietic cells. In addition to TCR and B cell antigen receptor engagement (3,9,11,12), ligand binding to transforming growth factor-␤ receptor (TGF-␤R) (13,14), the erythropoietin receptor (15), Fas (16), and E prostanoid receptors (17) can also induce HPK1 kinase activity. With the exception of TCR-mediated signal transduction where some mechanisms controlling HPK1 activation have been delineated, the exact biochemical mechanisms utilized by these receptors to activate HPK1 remain poorly understood. However, the signaling mechanisms utilized by these receptors can be grouped into three general categories: 1) Activation that depends on protein tyrosine kinases (PTKs), their substrates, and the Src homology (SH) domain-containing adapter proteins that couple these molecules to HPK1 (3,18,19); 2) Activation that relies on caspase activation and the subsequent cleavage of HPK1; 3) Activation that is presumed to utilize protein serine/threonine kinases (PS/TK) to conduct signals to HPK1 (13,14,17). Of these three categories, the PS/TK-dependent mechanism is the only mode of activation that has not been well characterized.
We recently reported that exposing T lineage or myeloid lineage cell lines to physiological concentrations of prostaglandin E 2 (PGE 2 ) would induce robust HPK1 kinase activity (17). Because the mechanisms that control PGE 2 -induced activation have not been delineated, we decided to characterize this signaling pathway by comparing it with the signaling mechanisms used by well characterized mechanism downstream of the TCR. lin/streptomycin). Lck-deficient JCaM 1.6 and the ZAP-70-deficient p116 Jurkat were obtained from ATTC (Mannasus, VA). The Lat-deficient ANJ Jurkat was a kind gift from Dr. Samelson (National Cancer Institute, National Institutes of Health), and the SLP-76-deficient J14 Jurkat was a kind gift from Dr. Koretzky (University of Philadelphia, PA). Wild type S49 pre T cells and the PKA-deficient mutant, kin Ϫ S49, were kind gifts from Dr. Insel (University of California, San Diego, CA).
Molecular Constructs-The pcDNA3.1 vector carrying the mouse cDNA that encodes the HA-tagged wild type murine HPK1 and the proline-rich-deleted construct (HA-HPK1 ⌬P) were created by Dr. F. Kiefer (11). PCR-assisted QuikChange mutagenesis system was used to introduce the desired point mutations (Stratagene, La Jolla, CA). Each mutated construct was sequenced to verify the existence of the desired mutation and for the absence of PCR-generated mutations.
Immunoprecipitations and Immunoblotting-Cells were lysed in a buffer containing 1% Nonidet P-40 and 50 mM Tris, pH 7.6, 150 mM NaCl, and a mixture of protease and phosphatase inhibitors. The lysates were pre-cleared with protein A-Sepharose, and subsequently proteins were immunoprecipitated with either 1 g of anti-human HPK1 no. 47 or 1 l of each of the anti-murine HPK1 nos. 5 and 6 antibodies or an equal amount of normal rabbit Ig by incubation at 4°C for 2 h. The beads were washed extensively with 0.1% Nonidet P-40 in immunoprecipitation wash buffer (150 mM NaCl, 20 mM HEPES, pH 7.4, 10% glycerol, 1 mM Na 3 V 4 O 7 , 5 mM NaF, 10 g/ml each of aprotinin and leupeptin), and the bead-bound proteins were either subjected to an in vitro kinase reaction (described below) or they were directly separated by SDS-PAGE. The proteins were transferred to a PVDF membrane, immunoblotted with the indicated antibodies, and developed by the enhanced chemiluminescence system (Amersham Biosciences).
Stimulation and in Vitro Kinase Assays-For stimulation by the anti-CD3⑀ antibody-mediated TCR cross-linking in suspension, Jurkat and its mutant lines were suspended in com-plete RPMI 1640 medium (1 ϫ 10 7 cells/immunoprecipitation) and incubated with 1 g of OKT3.14 anti-human CD3⑀ mAb at 4°C as described previously (20). After 10 min on ice with the stimulating antibodies, 7.5 g of rabbit anti-mouse antibody was added for cross-linking and incubated for an additional 10 min at 4°C. The cells were then warmed to 37°C for the indicated times prior to lysis. For stimulation with PGE 2 or cAMP analogues, these reagents were added to the 4°C cell suspension, and cells were immediately shifted to 37°C for 5 min to facilitate signal transduction. In the situation where H-89 PKA inhibitor was used to block PKA activation, S49 cells were pretreated with 10 M H-89 for 30 min prior to stimulation by PGE 2 . Whole cell lysates derived from resting or stimulated cells were subjected to immunoprecipitation by the indicated antibodies followed by in vitro kinase reactions as described (2). Cells were stimulated by cholera toxin for 30 min before whole cell lysate (WCL) is prepared. All stimulations were terminated by cell lysis, using a lysis buffer containing 1% Nonidet P-40, as described above.

PGE 2 -induced Activation of HPK1 Utilizes
Signaling Mechanism Distinct from that Used by the TCR-TCR signal transduction pathways leading to HPK1 activation are the best understood among all known receptors capable of activating HPK1. Studies using mutant cell lines lacking Lck and ZAP-70 revealed that the presence of these PTKs is required for HPK1 activation by the TCR (3). Because some rhodopsin-like G protein-coupled receptor use Lck to transduce signals via PTK-dependent pathways (21), we determined whether PGE 2 receptors in T cells must also engage these PTKs to activate HPK1 kinase activity. First, we assessed whether PGE 2 stimulation would induce general tyrosine phosphorylation in the Jurkat T cell line. Cells were left untreated or stimulated either with 10 nM PGE 2 or by anti-CD3⑀ (OKT3.14) mAb-mediated TCR cross-linking. Anti-phosphotyrosine immunoblotting of whole cell lysates revealed no detectable change in global tyrosine phosphorylation levels upon PGE 2 stimulation, whereas robust tyrosine phosphorylation was observed upon TCR cross-linking (Fig. 1A). Identical phosphorylation pattern was observed when cells were stimulated for varying time ranging from 2 to 10 min, but a modest tyrosine phosphorylation was detected at 20 min time point -a substantial amount of time after HPK1 was activated (supplemental Fig. S1, A and B). Anti-phosphotyrosine blotting of immunoprecipitated HPK1 confirmed that, unlike prominent tyrosine phosphorylation induced by TCR cross-linking (Fig. 1B, lane 3), stimulation by PGE 2 did not induce detectable tyrosine phosphorylation of HPK1 (lane 2). Western blot analysis using anti-human HPK1 antibody demonstrated that comparable amounts of HPK1 were present in all lanes (Fig. 1C). It also revealed that PGE 2 stimulation did not lead to activation-induced cleavage of HPK1, thus ruling out the involvement of caspase-mediated activation. The immune complex in vitro kinase (IVK) assay revealed that as in the previously reported studies (3,17), both TCR and PGE 2 receptors could activate HPK1 (Fig. 1D, lanes 2 and 3).
It has been shown that Jurkat somatic mutant cell lines that lack Lck and ZAP-70, J.CaM1 and p116, respectively, cannot activate HPK1 upon TCR engagement (3). Through the use of these mutant cell lines, we assessed whether HPK1 would catalytically respond to stimulation by PGE 2 . Wild type or mutant Jurkat cell lines were left untreated or stimulated with either 10 nM PGE 2 or by antibody-mediated TCR cross-linking. These cells were lysed, and the immunoprecipitated HPK1 was subjected to IVK analysis. We observed that the absence of Lck or ZAP-70 did not interfere with the ability of HPK1 to respond to PGE 2 stimulation ( Fig. 2A, lanes 5 and 8). As previously reported, the presence of these PTKs is required for an HPK1 response to TCR cross-linking ( Fig. 2A, lanes 6 and 9). Western blot analysis using an anti-HPK1 antibody indicated that comparable amounts of immunoprecipitated HPK1 were used in all IVK reactions (Fig. 2B, lanes 1-3 and 4 -9). We conclude from these studies that PGE 2 utilizes a PTK-independent pathway to induce HPK1 kinase activity.
HPK1 Is Responsive to PGE 2 Stimulation in Lat and SLP-76deficient Jurkat T Cell Lines-Scaffolding proteins play a critical role in transducing activation signals from the TCR to HPK1. It has been shown that the presence of Lat, and to a lesser extent SLP-76, is required for TCR-induced HPK1 activation (3). To assess the role of these scaffolding proteins in PGE 2 -induced HPK1 activation, we evaluated the mutant Jurkat T cell lines, ANJ3 and J14, which lacked the expression of Lat and SLP-76, respectively, for their ability to activate HPK1 in response to PGE 2 stimulation. A wild type Jurkat T cell line and the Lat and SLP-76 mutants were stimulated with 10 nM PGE 2 or by TCR cross-linking. Endogenous HPK1 was immunoprecipitated, and its catalytic activity was assessed by an IVK assay. Analysis of the receptor-induced HPK1 kinase activity revealed that all Jurkat cell lines could robustly induce HPK1 kinase activity upon PGE 2 stimulation ( Fig. 2A, lanes 2, 11, and 14), whereas the mutant cell lines failed to activate HPK1 in response to TCR engagement (lanes 12 and 15). Western blot analysis using an anti-HPK1 antibody indicated that comparable amounts of immunoprecipitated HPK1 were used in all IVK reactions (Fig.  2B, lanes 1-3 and 10 -15). We conclude that, unlike TCR-induced signaling to HPK1, Lat and SLP-76 are not involved in PGE 2 -induced HPK1 activation.
Proline-rich Regions of HPK1 Are Not Required for PGE 2 -induced HPK1 Activation-The interaction between SH3 domain-containing adapter proteins and the prolinerich motifs of HPK1 is critical for TCR-mediated signaling to HPK1 (3,8,9,11,(22)(23)(24). Three of the four proline-rich regions of HPK1 (P1, P2, and P4) conform to the class II consensus sequence for an SH3 protein interacting domain (11) and are known interaction sites for SH3 domain-containing proteins (25). To assess whether the proline-rich motifs in HPK1 are required in PGE 2 -induced HPK1 activation, we transfected constructs that encoded either the HA-tagged wild type HPK1 or a mutant form in which P1, P2, and P4 proline-rich motifs (HA-⌬P-HPK1) had been deleted. Trans-FIGURE 1. PGE 2 activates HPK1 without engaging the protein tyrosine kinase pathways. 20 million Jurkat cells were left untreated (Ϫ) or stimulated by 10 nM PGE 2 (P) or cross-linked by OKT3.14 mAb-mediated TCR crosslinking (CD3) at 37°C for 5 min. A, 100 g of WCL prepared from these cells were resolved electrophoretically by SDS-PAGE and Western blotted with the RC20H anti-phosphotyrosine mAb. B, half of the remaining lysates from A were subjected to immunoprecipitation by 1 g of an anti-human HPK1 polyclonal antibody 47 and Western blotted by the 4G10 anti-phosphotyrosine mAb. C, membrane from B was stripped, and the anti-human HPK1 polyclonal antibody 47 was used to determine the amount of immunoprecipitated HPK1 in each lane. D, HPK1 was immunoprecipitated from the remaining lysates from A and was subjected to an in vitro immune complex kinase assay. The kinase reaction was resolved electrophoretically by SDS-PAGE 12% gel, and the proteins were transferred to PVDF membrane. The autoradiographic bands depicted the 32 P-incorporated histone H2A catalyzed by HPK1 kinase activity. Data represent a reproducible trend observed in three out of three experiments. fectants were left untreated or stimulated with 10 nM PGE 2 or by TCR cross-linking. The ectopically expressed HPK1 proteins were immunoprecipitated with an anti-HA mAb and subjected to IVK assay. Analysis revealed that, whereas the HA-HPK1 ⌬P mutant was unable to respond to an TCR activation signal (Fig.  3A, lane 6), it responded robustly to PGE 2 stimulation (lane 5). This observation is consistent with the previous report that this mutant also failed to undergo TCR-induced tyrosine phosphorylation (11), an indication for TCR-induced HPK1 activation. HA-HPK1 responded to both forms of stimulation (Fig. 3A,  lanes 2 and 3), suggesting that the inability of the HA-HPK1 ⌬P mutant was not because of epitope tagging or the overexpression of HPK1. Western blot analysis indicated that comparable amounts of HPK1 were present in all IVK reactions (Fig. 3B). Thus, this finding suggests that P1, P2, and P4 proline-rich regions of HPK1 do not contribute toward the PGE 2 -induced HPK1 kinase activation, thus it is unlikely that adapter proteins play a role in this process.
Specific Serine/Threonine Phosphatase Inhibitors Activate HPK1 Kinase Activity-All E prostanoid receptors propagate PGE 2 -generated signals via PS/TK-dependent signal transduction pathways. To determine whether HPK1 kinase activity was regulated by PS/TK-dependent pathways, we compared HPK1 kinase activity from untreated Jurkat cells to those from cells treated with a panel of serine/threonine and tyrosine phosphatase inhibitors. Analysis of HPK1 IVK activity revealed that both calyculinA and okadaic acid, the inhibitors for protein phosphatase 1 (PP1) and 2A (PP2A), were able to activate HPK1 kinase activity (Fig. 4, lanes 2 and 3). Interestingly, cyclosporine A, the specific inhibitor of the TCR-responsive protein phosphatase 2B (calcineurin), failed to induce HPK1 kinase activity (lane 4). As reported previously (12), treating Jurkat cells with pervanadate, a tyrosine phosphatase inhibitor, induced a robust response in HPK1 catalytic activity (Fig. 4, lane 5). These data indicated that the catalytic activity of HPK1 is controlled by PP1-and PP2A-regulated PS/TK as well as by tyrosine kinases.
Elevation of cAMP Levels Induces HPK1 Activation-Prostaglandin E2 can bind with high affinity to each of the four E prostanoid receptors. However, EP2 and EP4, the receptors that couple with and signal through the stimulatory G␣ (G␣ S ) are the dominant receptors expressed in primary hematopoietic cells and hematopoietic cell lines (26 -29). The GTP-bound G␣ S subunit interacts with adenylyl cyclase and potentiates cAMP production upon PGE 2 stimulation. To test whether elevation of intracellular cAMP levels would activate HPK1 in Jurkat T cells, we stimulated cells with cholera toxin, cell permeable cAMP analogues, and forskolin, a potent adenylyl cyclase activator, to elevate cAMP levels and assess the catalytic activity of HPK1 in response to these stimulations. Analysis by an IVK assay revealed that, despite our initial belief that HPK1 might use cAMP-independent signaling pathways to activate HPK1, our current data showed that HPK1 responded to the stimulation by cholera toxin, N 6 , O 2Ј -dibutyryl-cAMP, 8-bromo-cAMP, and forskolin (Fig. 5A, lanes 2-5). Western blot analysis revealed that comparable amounts of HPK1 were present in all lanes (Fig. 5B).  Because PKA is the major effector molecule whose catalytic activity is regulated by the direct binding of cAMP, we tested whether PKA activity was required for PGE 2 -induced HPK1 activation. To test the involvement of PKA in this process, we stimulated the S49 pre-T cell line and its PKA-deficient mutant, the kin Ϫ S49 cell line, with PGE 2 and compared the ability of HPK1 to respond to PGE 2 stimulation. Similar to the response found in Jurkat cells, the catalytic activity of HPK1 in wild type S49 cells was elevated in response to stimulation by either PGE 2 or forskolin (Fig. 5C, lanes 2 and 3). The response of HPK1 could be blocked by pretreating the S49 cells 30 min before stimulation with 10 M H-89, an isoquinoline sulphoamide drug capable of specifically inhibiting PKA (Fig. 5C, lanes 4 and  5). In support of the critical role PKA plays in PGE 2 -induced HPK1 activation, the PKA-deficient kin Ϫ S49 mutant cell line was not able to activate HPK1 when stimulated by either PGE 2 or forskolin (Fig. 5C, lanes 7 and 8), confirming the critical role that PKA plays in this signaling pathway. Western blot analysis using an anti-HPK1 antibody indicated that comparable amounts of immunoprecipitated HPK1 were used in IVK reactions (Fig. 5D, lanes 1-8). To verify that HPK1 is catalytically active and is capable of responding to PKA-independent signals, we stimulated kin Ϫ S49 cells with pervanadate, a potent inhibitor of protein tyrosine phosphatase and a robust activator of HPK1. Our analysis indicated that HPK1 in kin Ϫ S49 cells responded catalytically to pervanadate (Fig. 5E), despite its inability to respond to PGE 2 stimulation (D). This finding suggests that HPK1 is catalytically active and is responsive to non-PKA-dependent signal transduction pathway. To further strengthen the evidence supporting the role of PKA in PGE 2induced activation of HPK1, we stimulated Jurkat cells with PGE 2 for 5 min, in the presence or absence of H-89, and Western blotted the immunoprecipitated HPK1 with anti-phospho PKA substrate antibody. Analysis revealed that PGE 2 -activated HPK1 was recognized by anti-phospho PKA substrate antibody (Fig. 5F). This recognition was abrogated by H-89 pre-treatment (Fig. 5F), suggesting that PKA directly phosphorylated HPK1. The dependence on PKA for PGE 2 -induced HPK1 activation sets it apart from TCR-induced HPK1 activation where PKA does not play a role in signal transduction (30).
Serine 171 Is Required for PGE 2 -induced HPK1 Activation-The susceptibility of PGE 2 -induced HPK1 activation to a PKA inhibitor, in conjunction with the inability of the PKA-deficient kin Ϫ S49 mutant cell line to activate HPK1 upon PGE 2 stimulation, strongly suggest that PKA is a critical upstream regulator of PGE 2 -induced HPK1 activation. These findings led us to analyze the HPK1 amino acid sequence for the presence of the optimal consensus PKA motif, the amino acid sequence RRXS/T, where X represents any amino acid (31). Sequence analysis identified serine 171, located within the "activation loop" of the kinase domain (the region flanked by the conserved "DFG" and "APE" amino acid sequences in the kinase subdomain VII and VIII), as the only optimal PKA site in HPK1 (Fig.   FIGURE 5. HPK1 is catalytically responsive to intracellular cAMP concentration. A, 10 million Jurkat cells were treated with the indicated cAMP-elevating agents at 37°C for 5 min (30 min for stimulation by cholera toxin) at the following concentrations: 10 ng/ml cholera toxin; 10 M N6; O 2 Ј-dibutyryl-cAMP; 10 M 8-bromo-cAMP; and 50 M forskolin. CTX, DB, 8BM, and Fors denote cholera toxin, N6, O 2 Ј-dibutyryl-cAMP, 8-bromo-cAMP, and forskolin, respectively. B, PVDF membrane containing the electrophoretically resolved proteins from IVK reactions was Western blotted with the anti-human HPK1 antibody 47. C, 10 million S49 cells or the PKAdeficient kin Ϫ S49 mutant line was stimulated with 10 nM PGE 2 or 50 M forskolin. S49 cells were pretreated with 10 M H-89 PKA inhibitor for 30 min at 37°C where indicated. Immunoprecipitated HPK1 isolated from these cells was subjected to an in vitro immune complex kinase assay. The autoradiographic bands depicted 32 P-incorporated histone H2A catalyzed by HPK1 kinase activity. D, PVDF membrane containing the resolved protein from IVK reaction was Western blotted with the anti-murine HPK1 antibody 7. E, Jurkat and kin Ϫ S49 mutant cells were stimulated with 10 nM PGE 2 (P) or 1 M pervanadate (Per) at 37°C for 5 min. Immunoprecipitated HPK1 isolated from these cells were subjected to an in vitro immune complex kinase assay. Data represent a reproducible trend observed in three out of three experiments. F, Jurkat cells were left unstimulated, stimulated by PGE 2 (P), or pretreated with 10 M H-89 PKA inhibitor for 30 min prior to stimulation by PGE 2 for 5 min at 37°C. 6A). Further analysis revealed that, whereas the arginine residue at Ϫ2 position relative to the serine 171 (arginine 169) was conserved in all KHS family members, only HPK1 possessed an arginine at the Ϫ3 (arginine 168) position relative to serine 171. The conserved double arginine sequence was also found in the murine HPK1 sequence, but not in the majority of Ste20 orthologues (data not shown). To assess the importance of arginine 168, arginine 169, and serine 171 in PGE 2 -induced HPK1 activation, we created a panel of mutant HPK1 expression constructs that encoded a point mutation at each of these amino acids by changing arginines to lysines and the serine to alanine. These mutant HPK1 constructs were transfected into Jurkat cells, and the ectopically expressed HA-tagged HPK1 were immunoprecipitated from resting or PGE 2 stimulated cells. Analysis of HPK1 kinase activity revealed that the serine 171 mutation (S171A) completely ablated the response of HPK1 to PGE 2 stimulation (Fig. 6B, lane 8). Arginine to lysine mutations at either residue 168 (R168K) or 169 (R169K) of HPK1 reduced the responsiveness to PGE 2 stimulation by ϳ90% (Fig. 6B, lanes  4 and 6). Western blot analysis using an anti-HPK1 antibody indicated that comparable amounts of immunoprecipitated HPK1 were used in the IVK reactions (Fig. 6C, lanes 1-8). This data suggest that the kinase, which phosphorylates serine 171 requires both arginine 168 and 169 to direct its substrate specificity.

DISCUSSION
We report here that TCR and PGE 2 receptors utilize distinct signaling mechanisms to activate HPK1. Whereas TCR relies on PTK-dependent signal transduction pathways and the intact proline-rich regions of HPK1 to transmit activation signals to HPK1, PGE 2 signals via PKA to activate HPK1. The reliance on PKA as the activator of HPK1 is consistent with the existing belief that PS/TK can signal to HPK1. This belief is based in part on the fact that ligand engagement of TGF-␤R, a receptor PS/TK, results in the activation of HPK1 (13,14). However, because TGF-␤ signaling pathways for HPK1 activation have not been characterized, the exact PS/TK pathway utilized by TGF-␤R to activate HPK1 remains unknown. With our observation that PKA plays a prominent role in PGE 2 signaling to HPK1, coupled with the recent report that TGF-␤ stimulation activates PKA in fetal skin fibroblasts (32), these findings support the possibility that the TGF-␤R may also use PKA to activate HPK1.
Catalytic activity of many kinases can be regulated by the phosphorylation of residues located within the activation loop (33,34). Structural conformation of the activation loop blocks ATP access to the catalytic pocket and thus prevents the induction of catalytic activity by steric hindrance. Upon phosphorylation by the upstream kinase cascade, the conformation of the activation loop is changed and moved away from the active site, which allows ATP free access to the kinase domain. The loss of the ability of HPK1 to respond catalytically to PGE 2 stimulation when serine 171 is mutated to alanine strengthens the hypothesis that PKA regulates HPK1 kinase activity by direct phosphorylation of the activation loop. The role of serine 171 in the regulation of HPK1 activity was further supported by a recent report that implicated PKD-dependent phosphorylation of serine 171 as one of the critical events necessary for TCR-induced activation of HPK1 (30). Because PKD is not activated by PGE 2 stimulation (supplemental Fig. S1, C and D), the requirement for serine 171 phosphorylation in two receptor systems that utilizes two different kinases highlights the importance of this phosphorylation in HPK1 activation. It is important to note here that, whereas our data and that of others indicate that serine 171 is indispensable for the activation of HPK1, we believe that the phosphorylation is a necessary, but not sufficient event that controls HPK1 kinase activity. Phosphorylation of immunoprecipitated HPK1 in vitro by purified PKA does not render HPK1 more catalytically active (data not shown). Thus, we conclude that other post-translational modifications, most likely additional phosphorylations by other PS/TK, are required to fully activate HPK1.
The PGE 2 -induced phosphorylation of HPK1's serine 171 represents a novel PKA-dependent activation mechanism not described previously for HPK1 or for other KHS family members. Among mammalian Ste20 orthologues, only MST-3B, the brain-specific splice variant isoform of MST3 has been previously shown to be catalytically responsive to PKA-mediated phosphorylation (35). In the case of MST-3B, however, the unique peptide segment that contains the PKA substrate motif vector alone or with the construct encoding the following HPK1 cDNAs: HAtagged wild type murine HPK1 (HA-HPK1), HA-tagged HPK1 mutant with serine 171 residue mutated to alanine (HA-HPK1 S171A), HA-tagged HPK1 mutants with either arginine 168 or 169 mutated to the lysine residue (HA-HPK1 R168K or HA-HPK1 R169K, respectively). Transfectants were stimulated with 10 nM PGE 2 at 37°C for 5 min. HPK1 was immunoprecipitated from lysates prepared from these cells, using the 12CA5 anti-HA epitope mAb, and was subjected to an in vitro immune complex kinase assay. The autoradiographic bands depicted 32 P-incorporated histone H2A catalyzed by HPK1 kinase activity. C, PVDF membrane containing the resolved protein from IVK reaction was Western blotted with the anti-murine HPK1 antibody 7. Data represent a reproducible trend observed in four out of six experiments.
is located N-terminal to the kinase domain. Thus, the inducible phosphorylation of HPK1's serine 171 represents the only report of a PKA-mediated inducible phosphorylation event that occurred within the activation loop of the mammalian Ste20 kinase domain. Perhaps this is a reflection of the infrequent occurrence of the double arginine motif in the activation loop of the Ste20 orthologues. Among the 31 human Ste20 orthologues identified by Sugen's genome analysis of human kinase genes, only MYO3, PAK4, SLK, and all MSN sub-family members (HGK, TNIK, MINK, and NRK) possess the double arginine motif at the location analogous to the double arginine residues of HPK1 (data not shown). We are currently investigating whether these kinases would respond catalytically to PKA.
Most tumors overexpress cyclooxygenase-2 and consequently produce high level of PGE 2 . Although PGE 2 stimulation favors tumor growth by activating the T cell factor/lymphocyte enhancer-binding factor signal transduction pathway in epithelial cells (36), it functions paradoxically as a potent inhibitor of T cell activation (37)(38)(39)(40). We propose here that HPK1, due in part to its hematopoietic cell-restricted expression pattern and responsiveness to PGE 2 , functions as a molecular switch that converts the pro-proliferation signal in epithelial cells into an inhibitory signal in hematopoietic cells. The fact that the PGE 2 -initiated signaling pathway converges on HPK1 despite the lack of overlapping membrane-proximal signaling mechanisms lends credibility to the proposal that PGE 2 produced by tumors may "hijack" the normal negative regulatory pathway mediated by HPK1 and use it to suppress the immune response against tumors. Such possibility, if verified, may lead to the development of a novel therapeutic approach that could overcome tumor-mediated immune suppression by inhibiting the catalytic activity of HPK1.